Module tests.test_cluster_autoscaler

Functions

def annotate_safe_to_evict_to_namespace_pods(core_api, namespace)

def configure_node_scale_down(core_api, nodes, disable)

def get_new_nodes(client, old_nodes)

def get_replica_count_to_scale_up(core_api, node_number, cpu_request)

def test_cluster_autoscaler(client, core_api, apps_api, make_deployment_cpu_request, request)

Scenario: Test CA

Given Cluster with Kubernetes cluster-autoscaler. And Longhorn installed. And Set kubernetes-cluster-autoscaler-enabled to true. And Create deployment with cpu request.

When Trigger CA to scale-up by increase deployment replicas. (double the node number, not including host node) Then Cluster should have double the node number.

When Trigger CA to scale-down by decrease deployment replicas. Then Cluster should scale-down to original node number.

def test_cluster_autoscaler_all_nodes_with_volume_replicas(client, core_api, apps_api, make_deployment_cpu_request, volume_name, pod_make, request)

Scenario: Test CA scale down all nodes with volume replicas

Given Cluster with Kubernetes cluster-autoscaler. And Longhorn installed. And Set kubernetes-cluster-autoscaler-enabled to true. And Create volume. And Attach the volume. And Write some data to volume. And Detach the volume. And Create deployment with cpu request.

When Trigger CA to scale-up by increase deployment replicas. (double the node number, not including host node) Then Cluster should have double the node number.

When Annotate new nodes with cluster-autoscaler.kubernetes.io/scale-down-disabled. (this ensures scale-down only the old nodes) And Trigger CA to scale-down by decrease deployment replicas. Then Cluster should have original node number + 1 blocked node.

When Attach the volume to a new node. This triggers replica rebuild. And Volume data should be the same. And Detach the volume. Then Cluster should scale-down to original node number. And Volume data should be the same.

def test_cluster_autoscaler_backing_image(client, core_api, apps_api, make_deployment_cpu_request, request)

Scenario: Test CA scale down with backing image

Given Cluster with Kubernetes cluster-autoscaler. And Longhorn installed. And Set kubernetes-cluster-autoscaler-enabled to true. And Create backing image. And Create deployment with cpu request.

When Trigger CA to scale-up by increase deployment replicas. (double the node number, not including host node) Then Cluster should have double the node number.

When Annotate new nodes with cluster-autoscaler.kubernetes.io/scale-down-disabled. (this ensures scale-down only the old nodes) And Trigger CA to scale-down by decrease deployment replicas. Then Cluster should have original node number + 1 blocked node.

When Remove backing image. Then Cluster should scale-down to original node number.

def wait_cluster_autoscale_down(client, core_api, nodes, diff)

def wait_cluster_autoscale_up(client, nodes, diff)